---
title: "Byte Latent Transformer: Patches Scale Better Than Tokens"
description: "BLT replaces a fixed subword vocabulary with entropy-driven byte patches. Follow the local encoder, global patch Transformer, fixed-inference scaling trade, character robustness gains, test-time threshold caveat, and the wall-clock evidence still missing."
canonical_url: "https://fanout.sh/daily/2026-11-21-byte-latent-transformer"
md_url: "https://fanout.sh/daily/2026-11-21-byte-latent-transformer.md"
last_updated: "2026-11-21"
access: "public"
---

# Byte Latent Transformer: Patches Scale Better Than Tokens

BLT replaces a fixed subword vocabulary with entropy-driven byte patches. Follow the local encoder, global patch Transformer, fixed-inference scaling trade, character robustness gains, test-time threshold caveat, and the wall-clock evidence still missing.

## Paper details

- Authors: Artidoro Pagnoni et al.

- Venue: ACL 2025

- Track: AI Research

- Difficulty: Hard

- Reading time: 12 min

- Original paper: https://aclanthology.org/2025.acl-long.453/

## What you will learn

- How next-byte entropy creates variable-length patches without assigning a learned token identity to each span.

- Why pooling bytes into fewer global positions can trade sequence work for a larger Transformer at fixed theoretical inference compute.

- What the matched-scale, fixed-inference, character, and noisy-text results establish, and where threshold changes and wall-clock costs narrow the claim.

## Continue exploring

[Inspect tokenizer boundaries](https://fanout.sh/labs/tokenizer-context): See how a fixed tokenizer changes sequence length and splits unfamiliar strings before comparing it with BLT's dynamic byte patches.

---
This representation contains public Fanout content only. Protected Pro lessons, account data, billing, checkout, and pricing are not included.

Browse the public content map: https://fanout.sh/sitemap.md
